Should satisfy EGI OLA (is there a global one or NGIs should follow their own?)

Existing grid sites

Virtual Machines in the repository. Manifest file should provide information about middleware version and in case of patched version the reason for patching i.e. what defects of middleware are being solved. This can be either bugs in the configuration procedure (e.g. missing files in a repository) or manual fixes in installed packages (e.g. update of kernel due to a security hole found).

Should be provided through a web based application

Quotas on users

“Charging” of users

Current limitations from IP range. Utilize VM-dedicated vlan in YPEPTH.

Bursting among StratusLab partner sites in case of pick loads. First priority before bursting on public clouds. First attempts for pan-european collaborative cloud.

Connectivity issues among TCD, CNRS and GRNET. How we can optimize network connectivity. Exploit QoS capabilities of GEANT.

Procedure to verify VMs produced internally in the project

Promote usage of VM creation recipe.

Enable download of VM images from AR web-front

VM signing and security.

Allow uploading of third-party images to StratusLab AR.

Verification process for third-party images. Processing of manifest files for verification of completeness. Automated verification process through Hudson (how we can check for misbehaving VMs?)

Cloud-admin monitoring console. Need to monitor the physical infrastructure, ONe services, VMs, grid sites on VMs.

User front-end. “Clients” should be able to monitor their allocated VMs. Access control is required.

VM profile used (memory size, CPUs, storage space)

Storage (EBS)

Network bandwidth (inbound and outbound)

Bursting to public clouds. Who decides it. How it is paid. How the user has control.

Pre-production cloud site.

Production cloud site to host stable releases

Update process for production services.

Availability zone (GRNET, CNRS)

Upgrade procedure. Migration of VMs.